Power distribution station room defect detection method and device based on insulating material decomposition product

Abstract

The invention discloses a power distribution station room defect detection method and device based on insulating material decomposition products. The method comprises the following steps: establishing simulation models respectively corresponding to a plurality of defects of the ring main unit in the power distribution station room; based on the simulation models corresponding to the multiple defects, composition characteristics corresponding to insulating material decomposition products corresponding to the multiple defects are determined; training a preset enhanced neural network model by using the composition features corresponding to the insulating material decomposition products corresponding to the plurality of defects to obtain a target recognition model; detecting a target component in the target power distribution station room and the concentration corresponding to the target component; and inputting the target component and the concentration corresponding to the target component into a target identification model, and determining a target defect corresponding to the target power distribution station room. The technical problem that the accuracy of the detected defect type is low due to the fact that the current partial discharge detection based on the electromagnetic and sound wave principle is influenced by factors such as a complex electromagnetic or sound source interference path on site is solved.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, and more specifically, to a method and apparatus for detecting defects in substation rooms based on the decomposition products of insulating materials. Background Technology

[0002] In the operation and maintenance of power systems, switchgear / ring mains units in substations play a crucial role. These devices are not only numerous but also widely distributed, serving as indispensable control and protection components in the power network. However, due to their complex internal structure and variable operating environment, problems such as partial discharge or overheating defects within the switchgear / ring mains units can often lead to serious consequences, including the tripping of entire distribution lines, severe overload of transfer lines, and even threats to the stability of the power system and the safety of users' electricity consumption.

[0003] Currently, the detection technology for internal defects in switchgear / ring main unit mainly relies on condition detection methods based on electromagnetic and acoustic principles. Partial discharge detection technology based on electromagnetic and acoustic principles is greatly affected by the complex electromagnetic environment or sound source interference on site, especially in places with complex electromagnetic environments such as power distribution rooms. This severely restricts the accuracy of detection, resulting in inaccurate detection of defect types and making it impossible to select appropriate measures for adjustment.

[0004] There is currently no effective solution to the above problems. Summary of the Invention

[0005] This invention provides a method and apparatus for detecting defects in substation rooms based on the decomposition products of insulating materials, which at least solves the technical problem that the accuracy of the detected defect types is low due to the influence of factors such as the complex electromagnetic or sound source interference paths in the field when using partial discharge detection based on electromagnetic and acoustic principles.

[0006] According to one aspect of the present invention, a method for detecting defects in a substation room based on the decomposition products of insulating materials is provided, comprising: establishing simulation models corresponding to multiple defects of a ring main unit in the substation room, wherein the multiple defects include partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials; determining the compositional features corresponding to the decomposition products of insulating materials corresponding to the multiple defects based on the simulation models corresponding to the multiple defects; training a preset enhanced neural network model using the compositional features corresponding to the decomposition products of insulating materials corresponding to the multiple defects to obtain a target recognition model; detecting the target components and their corresponding concentrations in the target substation room; and inputting the target components and their corresponding concentrations into the target recognition model to determine the target defects corresponding to the target substation room.

[0007] Optionally, simulation models corresponding to multiple defects of the ring main unit in the substation are established, including: establishing an initial simulation model corresponding to the ring main unit; determining the defect location and nature of each defect; setting the model parameters corresponding to each defect based on the defect location and nature of each defect; and adjusting the initial simulation model based on the model parameters corresponding to each defect to determine the simulation model corresponding to each defect.

[0008] Optionally, based on simulation models corresponding to multiple defects, the compositional characteristics of the insulation material decomposition products corresponding to each defect are determined, including: establishing a chemical reaction model based on the insulation material in the ring main unit; simulating environmental changes in the ring main unit under multiple defects based on the chemical reaction model and the simulation models corresponding to multiple defects, and determining the composition and concentration of the insulation material decomposition products corresponding to each defect at multiple time points; establishing a time series of concentration changes corresponding to each defect based on the composition and concentration of the insulation material decomposition products corresponding to each defect at multiple time points; and determining the compositional characteristics of the insulation material decomposition products corresponding to each defect based on the time series of concentration changes corresponding to each defect.

[0009] Optionally, a target recognition model is obtained by training a pre-defined enhanced neural network model using the compositional features of the insulation material decomposition products corresponding to each of the multiple defects. This includes: determining a training set based on the compositional features of the insulation material decomposition products corresponding to each of the multiple defects, wherein the training set includes the compositional features of the sample insulation material decomposition products and the actual defect types corresponding to the sample insulation material decomposition products; randomly initializing the classifier weights in the pre-defined enhanced neural network model; inputting the training set into the pre-defined enhanced neural network model to obtain the predicted defect types corresponding to the sample insulation material decomposition products; and adjusting the classifier weights of the pre-defined enhanced neural network model based on the predicted defect types and the actual defect types corresponding to the sample insulation material decomposition products to obtain the target recognition model.

[0010] Optionally, the target component and its corresponding concentration are input into the target identification model to determine the target defect corresponding to the target substation room. This includes: determining multiple combinations of insulation material decomposition products and their corresponding actual defect types based on the insulation material decomposition products corresponding to each defect; sequentially inputting the compositional features of the multiple combinations of insulation material decomposition products into the target identification model to obtain the predicted defect types corresponding to each of the multiple combinations of insulation material decomposition products; selecting the combination of insulation material decomposition products whose predicted defect type matches the actual defect type as the target sample combination; and inputting the component and its corresponding concentration that match the insulation material decomposition product type in the target sample combination into the target identification model to determine the target defect corresponding to the target substation room.

[0011] Optionally, the target defect degree corresponding to the target defect is determined based on the concentration of the target component; the target operation and maintenance method is determined based on the preset correspondence relationship, according to the target defect and the target defect degree, wherein the preset correspondence relationship characterizes the correspondence between the multiple defect degrees corresponding to multiple defects and the operation and maintenance methods.

[0012] According to another aspect of the present invention, a defect detection device for a substation room based on the decomposition products of insulating materials is also provided, comprising: a modeling module for modeling multiple defects of a ring main unit in the substation room, wherein the multiple defects include partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials; a determination module for determining the compositional features of the decomposition products of insulating materials corresponding to the multiple defects based on the simulation models corresponding to the multiple defects; a training module for training a preset enhanced neural network model using the compositional features of the decomposition products of insulating materials corresponding to the multiple defects to obtain a target recognition model; a detection module for detecting the target components and their corresponding concentrations in the target substation room; and an identification module for inputting the target components and their corresponding concentrations into the target identification model to determine the target defects corresponding to the target substation room.

[0013] According to another aspect of the present invention, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored program, wherein, when the program is running, the device where the non-volatile storage medium is located is controlled to execute any of the above-described methods for detecting defects in a substation based on the decomposition products of insulating materials.

[0014] According to another aspect of the present invention, a computer device is also provided, the computer device including a processor, the processor being configured to run a program, wherein the program, when running, executes any one of the above-described methods for detecting defects in a substation based on the decomposition products of insulating materials.

[0015] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements any of the above-described methods for detecting defects in a substation based on the decomposition products of insulating materials.

[0016] In this embodiment of the invention, a method for detecting defects in substation rooms based on the decomposition products of insulating materials is adopted. This involves establishing simulation models corresponding to multiple defects in the ring main unit within the substation room. These defects include partial discharge defects in metal components, surface discharge defects in insulators, air gap discharge defects in bushings, and overheating defects in insulating materials. Based on these simulation models, the compositional characteristics of the decomposition products of the insulating materials corresponding to each defect are determined. A pre-defined enhanced neural network model is trained using these compositional characteristics to obtain a target recognition model. The target components and their corresponding concentrations in the target substation room are detected. These target components and their corresponding concentrations are then input into the target recognition model to determine the target defects in the target substation room. This achieves the goal of identifying defects based on the decomposition products of insulating materials, thereby improving the accuracy of defect type identification. Furthermore, it solves the technical problem that current partial discharge detection methods based on electromagnetic and acoustic principles suffer from low accuracy due to complex electromagnetic or acoustic source interference paths in the field. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0018] Figure 1 A hardware block diagram of a computer terminal for implementing a method for detecting defects in a substation room based on the decomposition products of insulating materials is shown.

[0019] Figure 2 This is a flowchart illustrating a method for detecting defects in a power distribution room based on the decomposition products of insulating materials, according to an embodiment of the present invention.

[0020] Figure 3 This is a structural block diagram of a substation room defect detection device based on the decomposition products of insulating materials, provided in an embodiment of the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] According to an embodiment of the present invention, a method embodiment for detecting defects in a substation room based on the decomposition products of insulating materials is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal for implementing a method for detecting defects in substation rooms based on the decomposition products of insulating materials is shown. Figure 1 As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0025] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0026] The memory 104 can be used to store software programs and modules for application software, such as the program instructions / data storage device corresponding to the method for detecting defects in substation rooms based on the decomposition products of insulating materials in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the aforementioned application program for detecting defects in substation rooms based on the decomposition products of insulating materials. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0027] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0028] Figure 2 This is a flowchart illustrating a method for detecting defects in a substation based on the decomposition products of insulating materials, according to an embodiment of the present invention. Figure 2 As shown, the method includes the following steps:

[0029] Step S202: Establish simulation models corresponding to various defects of the ring main unit in the substation room. These defects include partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials.

[0030] This step involves establishing simulation models for various defects in the ring main unit of the substation. This includes mathematically describing the physical processes of each type of defect and using appropriate simulation software for model construction and operation. The model's geometry, physical parameters, and boundary conditions can be defined based on the defect type to be analyzed. The model region is divided into sufficiently small grids to ensure the accuracy and convergence of the simulation. Initial states and operating conditions for different defects are set based on experimental or field data, resulting in simulation models for each defect.

[0031] For example, partial discharge defects in metal components occur when sharp edges, protrusions, or burrs exist on metal parts inside electrical equipment. The electric field around these areas becomes uneven, especially under high voltage conditions, potentially leading to partial discharge. Partial discharge is a discharge occurring within a specific area of ​​the insulating material under high voltage, but it does not immediately cause insulation breakdown. It can gradually damage the insulating material, reducing its performance and eventually leading to complete insulation failure. These defects are typically identified using partial discharge detection techniques such as UHF and ultrasonic testing (US). Creating a simulation model of partial discharge defects in metal components involves first creating a 3D model containing the metal component and insulating material, noting the location and shape of protrusions or burrs, as well as the electrode arrangement. Then, the conductivity of the metal component, the dielectric constant and loss factor of the insulating material, and the voltage conditions between the electrodes are set. Electromagnetic field simulation software is then used to simulate the electric field distribution and partial discharge phenomenon.

[0032] Surface discharge defects in insulators occur on the surface of the insulator. When the insulator is contaminated (e.g., by dust, salt) or humidity increases, its surface resistance decreases. Combined with the effect of an electric field, this can lead to partial discharge on the surface, i.e., surface discharge. This discharge phenomenon can further evolve into arc discharge or flashover, seriously affecting the reliability of the insulator and even causing the failure of the entire equipment. Surface discharge detection typically combines infrared thermal imaging, ultraviolet imaging, and partial discharge detection techniques. To create a simulation model of surface discharge defects in insulators, a geometric model of the insulator can be created first, including its complex surface shape and possible contamination layers. Then, the dielectric properties of the insulator are set, considering the impact of surface contamination on conductivity and electrochemical reactions under humidity conditions. Finally, simulation software for electromagnetic fields and heat and mass transfer is used to simulate the surface discharge process and the generated gases.

[0033] Bushing air gap discharge defects refer to the presence of air bubbles in the liquid or gas insulating bushings of electrical equipment (such as transformers) during manufacturing or assembly. These bubbles can become air gaps under high voltage or high field strength conditions, leading to partial discharge. Air gap discharge occurs when the air inside the air gap breaks down, generating an electric arc, consuming energy, and potentially causing localized damage to the bushing material, thus affecting insulation performance. Detecting such defects typically requires partial discharge detection technology and high-frequency measurement technology. Creating a simulation model of bushing air gap discharge defects involves first constructing a model of the air gap inside the bushing, including its size, location, and shape, as well as the geometry of the surrounding bushing material. The conductivity, dielectric constant, and breakdown voltage of the gas within the air gap under high field strength are then set. Finally, simulation software is used to create a composite model combining electromagnetic fields and gas discharge.

[0034] Overheating defects in insulation materials typically occur in areas of poor contact, excessive current, or inadequate heat dissipation within equipment. When current flows through a conductor or contact surface, if resistance increases, excess electrical energy is converted into heat, leading to a localized temperature rise—this is called overheating. High temperatures can accelerate the aging of insulation materials, reduce their electrical performance, and prolonged overheating can even cause serious accidents such as short circuits and fires. Common techniques for monitoring and diagnosing overheating defects include infrared thermal imaging detection and temperature sensor monitoring, as well as analyzing the gaseous components produced by the decomposition of insulation materials. Creating a simulation model of overheating defects in insulation materials involves first constructing a model of the conductor connection, including the contact surface and the geometry of the insulation material. The thermal conductivity, specific heat capacity, contact resistance of the material, and the thermal boundary conditions of the surrounding environment are then set. Thermal simulation software is used to simulate the temperature distribution and overheating phenomena when current flows through it.

[0035] Through the construction and analysis of the above simulation models, we can gain a deeper understanding of the physical processes under different defect conditions, providing a theoretical basis and data support for the development of fault detection methods and operating strategies based on the decomposition products of insulating materials.

[0036] Step S204: Based on the simulation models corresponding to each of the multiple defects, determine the compositional characteristics of the decomposition products of the insulating material corresponding to each of the multiple defects.

[0037] In this step, gaseous components (O3, NO) are collected from the simulation model. nConcentration data of gases (such as CO, etc.) at different time points and locations are collected. The collected data ensures coverage of the complete partial discharge or overheating process, including the initiation, development, and stabilization phases. For each type of defect, the trends in gas composition over time are analyzed. This may include: concentration fluctuations, i.e., observing the upward trend of a specific gas concentration at the time of defect occurrence; generation rate, to understand the gas generation dynamics during defect development; and steady state, i.e., determining the point where the gas concentration reaches a steady state and analyzing the gas composition under steady-state conditions. Spatial gas distribution can also be analyzed, including gas distribution patterns, i.e., determining the diffusion patterns of gases within the device, which helps identify the location of defects. Hotspot analysis may also be included, i.e., for overheating defects, identifying gas composition changes in the highest temperature regions, which may generate more characteristic gases. Local accumulation may also be included, i.e., assessing whether gases accumulate in certain local areas, which may be due to gas generation sources or diffusion barriers. Time-series and spatial distribution data can be fused to construct a feature vector. The feature vector may include: gas type, concentration changes, time dependence, location information, etc.

[0038] Furthermore, based on the above, a database containing decomposition products of insulating materials with different defect types can be constructed. The database should record detailed information such as the type, concentration, generation time, and location distribution of characteristic gases for each defect type. Subsequent model training can then extract and select the gas composition characteristics that best represent each defect type from the database.

[0039] Step S206: The pre-set enhanced neural network model is trained using the compositional features of the decomposition products of the insulating material corresponding to each of the multiple defects to obtain the target recognition model.

[0040] In this step, the compositional features of the decomposition products of the insulating material corresponding to each of the multiple defects are used to train a pre-defined enhanced neural network (e.g., BP-AdaBoost) to obtain a target recognition model capable of accurately identifying different defect types. Gas composition data (O3, NO) collected from simulation models or experiments can be organized and prepared. nThe dataset is prepared by dividing the input gas composition data into training and testing sets. The training set is used to learn the model, and the testing set is used to verify the model's generalization ability. Typically, the ratio is 80% training and 20% testing, or cross-validation can be used. The BP neural network is initialized, including setting the network structure (number of neurons in the input, hidden, and output layers), activation functions (e.g., sigmoid, ReLU), optimizers (e.g., gradient descent, Adam), and loss functions (e.g., mean squared error (MSE), cross-entropy). Based on the BP neural network, the AdaBoost algorithm is used for ensemble learning. AdaBoost creates multiple weak classifiers (each in a BP neural network) and then combines them into a strong classifier through weighted voting to improve the model's recognition accuracy and robustness. The BP-AdaBoost model is trained using the training set data, and the weights and thresholds are iteratively optimized to enable the model to predict the correct defect type based on the input gas composition data. The training can terminate when a preset number of iterations is reached, or when the model's loss function value falls below a preset threshold.

[0041] During training, the model's recognition performance is optimized by adjusting the network structure, parameters (such as learning rate and regularization), and hyperparameters (such as the number of AdaBoost iterations and the number of weak classifiers). The model's recognition accuracy and generalization ability can be validated using test set data. The model's prediction accuracy on the test set is calculated and compared with other techniques or models to evaluate the performance of the BP-AdaBoost model.

[0042] Once the model has been trained and its performance verified to be reliable, it can be deployed in a real-world substation ring main unit condition monitoring system. This system can then be used to detect insulation material decomposition products in real-time or periodically, automatically identify defect types, and provide corresponding alarms or maintenance recommendations. Through these steps, a BP-AdaBoost target recognition model based on the characteristics of insulation material decomposition products can be trained. This model can accurately classify and identify typical defects in ring main units in substations, providing intelligent decision support for power system condition monitoring and maintenance.

[0043] Step S208: Detect the target component and its corresponding concentration in the target power distribution room.

[0044] In this step, gas detection points can be set at different key locations in the target substation to obtain the overall distribution of the decomposition products of the insulation material, such as near insulators, cable troughs, and ventilation openings. A reasonable sampling frequency can be set according to the gas generation rate and detection requirements to ensure that changes in gas concentration can be captured. Using the selected detection instrument, the target components are detected according to the preset procedure. Ensure complete data recording, including time and concentration information.

[0045] Step S210: Input the target component and its corresponding concentration into the target identification model to determine the target defect corresponding to the target substation room.

[0046] In this step, the concentration data of the gas components is used as the input vector and input into the target recognition model according to the model's design format. The input data should contain the concentration information of all characteristic gases learned by the model during the training phase. The target recognition model is then run to process the input gas concentration data. Through the model's internal neural network and AdaBoost algorithm, the corresponding defect type is predicted. The model's output may be a classification label or a probability distribution for each defect type. Based on the model's output, it is determined whether a specific defect type exists in the substation, such as partial discharge of metal components, surface discharge of insulators, air gap discharge of bushings, or overheating of insulation materials. The model's prediction results can help maintenance personnel quickly locate potential problems and determine the next course of action.

[0047] A detailed inspection report can be prepared based on the defect type predicted by the model. The report should include the characteristic gas detected, its concentration, the predicted defect type, and the recommended maintenance or repair measures.

[0048] Through the above steps, the status of ring main units in substations can be assessed in real time or periodically using the BP-AdaBoost model based on gas derivative detection. By analyzing the gas composition characteristics and concentrations of the decomposition products of insulation materials, potential defect types can be automatically identified, providing intelligent support for the condition monitoring and maintenance of power systems. This method can overcome the shortcomings of traditional detection methods, especially in detecting early and intermittent defects, where it has significant advantages.

[0049] Through the above steps, the goal of identifying defects based on the decomposition products of insulating materials can be achieved, thereby improving the accuracy of defect type identification. This solves the technical problem that the accuracy of defect type detection is low due to the influence of factors such as complex electromagnetic or acoustic source interference paths in the field, which currently affect partial discharge detection based on electromagnetic and acoustic principles.

[0050] As an optional embodiment, a simulation model is established for each of the multiple defects of the ring main unit in the substation, including: establishing an initial simulation model for the ring main unit; determining the location and nature of each of the multiple defects; setting the model parameters for each of the multiple defects based on the location and nature of each of the multiple defects; and adjusting the initial simulation model based on the model parameters for each of the multiple defects to determine the simulation model for each of the multiple defects.

[0051] Optionally, an initial simulation model of the ring main unit can be established. This can be done using 3D modeling software to create a geometric model of the ring main unit, including key components such as metal parts, insulators, bushings, and cable connectors. Then, realistic physical properties are assigned to the model objects, such as the conductivity and magnetism of the metal parts, and the dielectric constant, loss factor, and heat resistance of the insulating materials. The simulation environment conditions, including temperature, humidity, and atmospheric pressure, as well as potential sources of electromagnetic interference, are set.

[0052] Determine the location and nature of different defects, for example:

[0053] Partial discharge defects in metal components: Simulate the location of protrusions or burrs on metal components, set the conditions for connection to a high-voltage power supply, and the distance from ground.

[0054] Insulator surface discharge defects: Determine the location of the insulator, consider the impact of surface contamination and moisture on insulation performance, and determine the connection method between high voltage and ground.

[0055] Bushing air gap discharge defects: Simulate the presence of internal air bubbles, set the bushing material and air gap location, and the contact between the high-voltage electrode and the bushing.

[0056] Insulation material overheating defects: Determine the location of wire connection, set contact resistance and current load, and consider heat dissipation path and ambient temperature.

[0057] Then, model parameters are set, which can include electrical, thermal, and chemical parameters. Electrical parameters can include voltage, current, electric field strength, and defect-related electrical parameters (such as contact resistance and breakdown voltage). Thermal parameters can include the material's thermal conductivity, specific heat capacity, and the power and heat dissipation conditions of the heat source. Chemical parameters can include chemical reaction rates related to gas generation and decomposition temperature thresholds. Then, based on the determined defect location and properties, and the model parameters, the details of the initial simulation model are adjusted to ensure that the simulation results accurately reflect the situation under the defect.

[0058] Through this series of steps, a sophisticated simulation model can be established for each type of defect. These models can be run under different conditions to study the generation law and distribution characteristics of the decomposition products of insulating materials, thereby providing theoretical basis and technical support for the condition detection and maintenance of the ring main unit in the substation.

[0059] As an optional embodiment, based on simulation models corresponding to multiple defects, the compositional characteristics of the insulation material decomposition products corresponding to each defect are determined, including: establishing a chemical reaction model based on the insulation material in the ring main unit; simulating environmental changes in the ring main unit under multiple defects based on the chemical reaction model and the simulation models corresponding to multiple defects, and determining the composition and concentration of the insulation material decomposition products corresponding to each defect at multiple time points; establishing a time series of concentration changes corresponding to each defect based on the composition and concentration of the insulation material decomposition products corresponding to each defect at multiple time points; and determining the compositional characteristics of the insulation material decomposition products corresponding to each defect based on the time series of concentration changes corresponding to each defect.

[0060] Optionally, based on the insulation material in the ring main unit, a chemical reaction model is established. Using this model and simulation models for different defects, the environmental changes within the ring main unit under various defects are simulated to determine the composition and concentration of the insulation material decomposition products. Furthermore, a time series of concentration changes is established to determine the compositional characteristics corresponding to different defects. This allows for the acquisition of the chemical reaction mechanism of insulation material decomposition under partial discharge or overheating conditions, identifying key reaction pathways and the types of gases generated. Kinetic parameters of the chemical reaction, such as reaction rate constant, activation energy, and decomposition temperature, are determined. A chemical reaction model is established using chemical software and then integrated into the simulation model for each defect, ensuring that the generation and consumption of insulation material decomposition products can be calculated in real time during the simulation. The coupling relationship between electrochemical, thermochemical, and physical conditions is established to ensure that the model can reflect actual environmental changes. Reasonable boundary conditions, such as initial gas concentration, pressure, temperature, and flow rate, are set to simulate the real internal environment of the ring main unit. The model is run, considering the interactions of electricity, heat, and chemistry. During the simulation, the composition and concentration data of insulation material decomposition products at multiple time points under different defect conditions are recorded. The collected component and concentration data are arranged according to time series to form a time series dataset. Features are extracted from the time series data, such as the trend of gas concentration changes, the time of peak occurrence, and the correlation between different gas components.

[0061] Statistical analysis or machine learning methods, such as principal component analysis (PCA), cluster analysis, support vector machine (SVM), and deep learning, can be used to analyze time-series data of insulation material decomposition products under different defects. The most representative gas components and their concentration variations under each defect condition are identified; these become the compositional features for identifying different defects. The identified compositional features are then compiled and stored in a database to construct a database of insulation material decomposition products under different defect types for subsequent analysis and model training.

[0062] Specifically, a characteristic gas spectrum library was constructed based on simulation tests and physical switchgear tests for different defect types, defect locations, detection locations, and environmental factors. The composition and content of characteristic components under different operating conditions were statistically analyzed, and a total of four types of defects were statistically analyzed. Among them, the characteristic gas spectrum library for insulation defects has the following characteristics:

[0063] (1) The gas composition under partial discharge defects is NO2, CO and O3, and the gas composition under local overheating defects is CO and O3.

[0064] (2) The characteristic gases under partial discharge defects are consistent in type, and their relative contents show no obvious pattern. Under most operating conditions, the relative content of NO2 is greater than that of O3, i.e. (NO2)> The content of (O3) and CO varies greatly with environmental factors and the intensity of discharge.

[0065] (3) The data characteristics of different defects under partial discharge conditions are not obvious, and it is impossible to effectively distinguish various types of discharge defects.

[0066] (4) The characteristic gas composition under local overheating defects is consistent, and the relative content is: (CO)> (O3), and the more severe the defect, the higher the content of the component.

[0067] (5) Under the same material, when the degree of overheating defects is consistent, the material with poorer heat resistance will generate more characteristic components under defects.

[0068] Through the above steps, a systematic model can be established for the generation of decomposition products of insulation materials in ring main units under different defect conditions, as well as the characteristic sequences of the composition and concentration of these products over time. This provides high-quality training data for subsequent intelligent identification algorithms based on gas derivatives. The established model and feature library will serve as important tools for power system condition monitoring and fault diagnosis, improving detection efficiency and accuracy.

[0069] As an optional embodiment, a target recognition model is obtained by training a preset enhanced neural network model using the compositional features of the insulation material decomposition products corresponding to multiple defects. This includes: determining a training set based on the compositional features of the insulation material decomposition products corresponding to multiple defects, wherein the training set includes the compositional features of the sample insulation material decomposition products and the actual defect types corresponding to the sample insulation material decomposition products; randomly initializing the classifier weights in the preset enhanced neural network model; inputting the training set into the preset enhanced neural network model to obtain the predicted defect types corresponding to the sample insulation material decomposition products; and adjusting the classifier weights of the preset enhanced neural network model based on the predicted defect types and the actual defect types corresponding to the sample insulation material decomposition products to obtain the target recognition model.

[0070] Optionally, based on the compositional characteristics of the insulation material decomposition products corresponding to multiple defects, a pre-defined enhanced neural network model (e.g., BP-AdaBoost model) is established and trained to obtain a target recognition model capable of accurately identifying defect types. The composition and concentration of insulation material decomposition products obtained from experiments and simulations, along with the corresponding defect types (e.g., partial discharge of metal components, surface discharge of insulators, air gap discharge of bushings, and overheating of insulation materials), are organized to ensure that each data point contains complete gas composition information and a clear defect type label. Each record in the training set is ensured to include two parts: a set of features (i.e., gas composition and concentration) and the corresponding defect type label; this is the basis of supervised learning. In the BP neural network, the connection weights and thresholds between all neurons are randomly initialized. For the AdaBoost ensemble learning algorithm, the weights of each weak classifier (BP neural network) also need to be initialized; typically, the initial weights of each weak classifier are the same and are adjusted during iteration. The data in the training set (compositional characteristics of insulation material decomposition products) is input into the BP neural network, and forward propagation begins, calculating the predicted defect type of the output layer. The predicted defect type corresponding to the sample insulation material decomposition products obtained after each forward propagation is recorded. Calculating the error between the predicted defect type and the actual defect type is the basis of backpropagation. The weights and thresholds in the neural network are adjusted backward based on the error to minimize the prediction error. For each training iteration, the weights of the BP neural network in AdaBoost are adjusted according to its prediction error rate; the weights of networks with high error rates are decreased, and vice versa. These steps are repeated until the network converges, i.e., the prediction error is minimized or after a predetermined number of iterations. In the BP-AdaBoost algorithm, all the weak classifiers (BP neural networks) obtained after training are integrated into a strong classifier to form the final target recognition model. The target recognition model is evaluated using test set data that was not used in training to ensure its generalization ability on unknown data.

[0071] Through the above process, a target recognition model based on a database of insulation material decomposition products and an enhanced neural network model can be trained. This model can predict the specific defect types present in the ring main unit of the substation based on the input gas component concentration. In practical applications, continuous optimization and updating of the model are also necessary to adapt to new data and continuously improve the recognition accuracy.

[0072] As an optional embodiment, the target component and its corresponding concentration are input into the target identification model to determine the target defect corresponding to the target substation room. This includes: determining multiple combinations of sample insulation material decomposition products and their corresponding actual defect types based on the insulation material decomposition products corresponding to each defect; sequentially inputting the compositional features of the multiple sample insulation material decomposition product combinations into the target identification model to obtain the predicted defect types corresponding to each of the multiple sample insulation material decomposition product combinations; selecting the sample insulation material decomposition product combinations whose predicted defect types match the actual defect types as target sample combinations; and inputting the target component and its corresponding concentration that match the insulation material decomposition product types in the target sample combinations into the target identification model to determine the target defect corresponding to the target substation room.

[0073] Optionally, organize the data on insulation material decomposition products obtained from previous experiments and simulations, and identify multiple gas combination samples. Each sample should contain a specific set of insulation material decomposition products and their concentrations. Label each sample with its corresponding actual defect type, establishing a sample library where each sample is associated with actual defect type information. Input each sample combination (i.e., gas components and concentrations) from the sample library into the target recognition model. Obtain the defect type predicted by the model for each sample combination and compare it with the actual defect type of the sample. Retain sample combinations whose model predictions perfectly match the actual defect types; these combinations will serve as reference standards for subsequent model applications. In the target substation room, use gas detection equipment to perform on-site detection of insulation material decomposition products, obtaining the target components and their corresponding concentrations. Compare the actually detected target components with the component types in the target sample combinations to find possible matches. Input the target components and their concentration data that match the component types in the target sample combinations into the target recognition model. The model will predict the possible defect types in the target substation room based on the input gas characteristics.

[0074] For example, if the detection results are relatively accurate when the target sample combination is found to contain both NO2 and O3, then the gas concentrations corresponding to NO2 and O3 in the target component can be input into the target recognition model for defect detection.

[0075] This process fully leverages the accuracy of model identification and the real-time nature of on-site detection data, providing efficient and reliable decision support for the operation and maintenance of substations. For example, if a significant increase in NO2 and O3 concentrations is detected on-site, and the model demonstrates high accuracy in predicting sample combinations containing these two gases during the validation phase, the NO2 and O3 concentration data can be input into the model to quickly locate potential partial discharge or overheating defects in metal components within the substation. This allows for targeted maintenance measures to prevent the fault from escalating and improve the safe operation of the power grid. Through continuous feedback of on-site detection data and iterative model optimization, this system can be continuously improved in practice, enhancing its predictive accuracy and practicality.

[0076] This process integrates preliminary experimental and simulation data with on-site inspection results, utilizing a target recognition model for intelligent defect type identification, providing a scientific basis for the maintenance and fault prevention of substations. The model's validation and application steps ensure its accuracy and practicality, while also highlighting the unique value of the detection method based on an insulation material decomposition product database in power equipment condition monitoring. Through continuous data collection and model optimization, the accuracy and efficiency of detection can be further improved, providing more reliable and intelligent maintenance strategies for power systems.

[0077] As an optional embodiment, the target defect degree corresponding to the target defect is determined based on the concentration of the target component; based on the preset correspondence, the target operation and maintenance method is determined according to the target defect and the target defect degree, wherein the preset correspondence represents the correspondence between the multiple defect degrees corresponding to multiple defects and the operation and maintenance methods.

[0078] Optionally, the concentration of the target component detected from the target substation can be analyzed and compared with typical concentration ranges for different defect types in experimental and simulation databases. Based on the gas concentration, combined with experimental data and industry standards, the target defect is classified into different severity levels, such as minor, moderate, and severe. A pre-defined correspondence table is created, which should list in detail the recommended operation and maintenance methods for different defect types at each severity level. For example:

[0079] Mild: Record and continuously monitor, and periodically review gas concentrations.

[0080] General: Perform local maintenance and inspection, such as tightening connections and cleaning the surface of insulators.

[0081] Serious: Immediately shut off the power and conduct a detailed inspection; replace equipment parts if necessary.

[0082] Based on the target defect type and the determined defect severity, the corresponding operation and maintenance method is searched in the preset correspondence table. Following the found operation and maintenance method, the target substation is immediately maintained. For example, if the model predicts a severe partial discharge defect with metal protrusions, maintenance personnel should immediately arrange a power outage for inspection and consider replacing or repairing the defective metal components.

[0083] Specifically, the ring main unit status can be classified into the following levels based on the above test results. Four different maintenance strategies (AD and A) are adopted for different defect degrees, each containing multiple countermeasures. The operation and maintenance methods can be selected according to the following correspondence. It should be noted that the evaluation criteria for each gas component do not need to be strictly met, nor do the contents of all three gases need to be satisfied. When a single gas component reaches the warning value, corresponding measures should be taken.

[0084] The corresponding relationship can be: when no characteristic components are detected, there are no defects, it is a normal state, and operation can continue. When NO is detected... n If CO and O3 are present and their maximum concentration does not exceed 10 μL / L, the defect type is considered a partial discharge defect, and the degree of defect is considered general. A combined TEV and AE method should be used for diagnosis. If CO and O3 are present and their maximum concentration does not exceed 2 μL / L, the defect type is considered an overheating defect, and the degree of defect is considered general. A combined infrared thermography method should be used for diagnosis. If CO and O3 are present and their maximum concentration exceeds 10 μL / L, the defect type is considered an overheating defect, and the degree of defect is considered severe. Power outage and inspection of the abnormal temperature rise location are required. If NO2, CO, and O3 are present and their maximum concentration exceeds 10 μL / L, the defect type is considered a partial discharge defect, and the degree of defect is considered severe. Power outage, equipment replacement, and disassembly analysis are required. TEV refers to the conduction effect of partial discharge on the casing of electrical equipment. When partial discharge occurs in the insulating medium, high-frequency current pulses are generated. These pulses propagate along the metal casing and generate electromagnetic radiation and electrical vibration signals at grounding points or discontinuities. TEV (Transient Emission Vehicle) technology locates and assesses partial discharges by detecting these electrical vibration signals. AE (Acoustic Emission) technology is based on the principle that partial discharge activity releases sound waves or stress waves. When a partial discharge occurs, a portion of the energy is converted into acoustic or mechanical energy, propagating through the metal casing into the surrounding space. AE sensors capture these sound waves or stress waves, and then analyze the characteristics of the partial discharge. In the condition monitoring of power equipment, TEV and AE methods are often used in combination to provide more comprehensive and accurate partial discharge detection and assessment. The complementarity of the two methods increases the reliability and coverage of detection, especially in complex environments and equipment, enabling more effective location and identification of potential partial discharge defects.

[0085] The analysis of characteristic gas component detection results can also employ longitudinal analysis and cross-sectional analysis methods. Generally, if the difference between the detection result and the environmental background value is greater than 0 ppm, the cause needs to be investigated. The environmental background value refers to the gas component and concentration values ​​measured by the gas detection equipment when there are no partial discharges, overheating, or other types of defects inside the equipment. When the content of a certain component exceeds 2 ppm, a relatively obvious discharge phenomenon exists. If necessary, other detection methods such as TEV and AE methods should be used to assist in the judgment.

[0086] By determining the defect severity based on the target component concentration and combining it with a pre-defined operation and maintenance correspondence table, rapid response and precise handling of defects in the target substation can be achieved. This method not only improves maintenance efficiency but also reduces unnecessary equipment downtime, significantly positively impacting the reliability and economy of the power system. Importantly, this process should be a closed-loop system, continuously collecting field data and feedback maintenance results to constantly adjust and optimize the pre-defined correspondence table, ensuring the optimization of the operation and maintenance strategy and the long-term effectiveness of the model.

[0087] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that the substation defect detection method based on the decomposition products of insulating materials according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0089] According to embodiments of the present invention, a substation room defect detection device based on insulation material decomposition products is also provided for implementing the above-described method for detecting defects in substation rooms based on insulation material decomposition products. Figure 3 This is a structural block diagram of a substation room defect detection device based on the decomposition products of insulating materials according to an embodiment of the present invention, as shown below. Figure 3As shown, the substation room defect detection device based on insulation material decomposition products includes: a setup module 302, a determination module 304, a training module 306, a detection module 308, and an identification module 310. The substation room defect detection device based on insulation material decomposition products will be described below.

[0090] Module 302 is used to establish simulation models corresponding to various defects of the ring main unit in the substation room. These defects include partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials.

[0091] The determination module 304, connected to the establishment module 302, is used to determine the composition characteristics of the decomposition products of the insulating material corresponding to each of the multiple defects based on the simulation models corresponding to each of the multiple defects.

[0092] The training module 306, connected to the determination module 304, is used to train a preset enhanced neural network model using the compositional features of the decomposition products of the insulating material corresponding to multiple defects, so as to obtain a target recognition model.

[0093] The detection module 308, connected to the training module 306, is used to detect the target components and their corresponding concentrations in the target power distribution room.

[0094] The identification module 310 is used to input the target component and the concentration corresponding to the target component into the target identification model to determine the target defect corresponding to the target substation room.

[0095] It should be noted that the aforementioned establishment module 302, determination module 304, training module 306, detection module 308, and recognition module 310 correspond to steps S202 to S210 in the embodiments. Multiple modules implement the same instances and application scenarios as their corresponding steps, but are not limited to the content disclosed in the above embodiments. It should also be noted that the above modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0096] Embodiments of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0097] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the method and device for detecting defects in substation rooms based on the decomposition products of insulating materials in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned method for detecting defects in substation rooms based on the decomposition products of insulating materials. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0098] The processor can access the information and application programs stored in the memory via a transmission device to execute the following steps: Establish simulation models corresponding to multiple defects in the ring main unit of the substation, including partial discharge defects in metal components, surface discharge defects in insulators, air gap discharge defects in bushings, and overheating defects in insulating materials; Based on the simulation models corresponding to the multiple defects, determine the compositional characteristics of the decomposition products of the insulating materials corresponding to each defect; Train a pre-defined enhanced neural network model using the compositional characteristics of the decomposition products of the insulating materials corresponding to the multiple defects to obtain a target recognition model; Detect the target components and their corresponding concentrations in the target substation; Input the target components and their corresponding concentrations into the target recognition model to determine the target defects corresponding to the target substation.

[0099] Optionally, the processor may also execute program code for the following steps: establishing simulation models corresponding to multiple defects of the ring main unit in the substation, including: establishing an initial simulation model corresponding to the ring main unit; determining the defect location and nature of each defect; setting model parameters corresponding to each defect based on the defect location and nature of each defect; adjusting the initial simulation model based on the model parameters of each defect, and determining the simulation model corresponding to each defect.

[0100] Optionally, the processor may also execute program code for the following steps: based on simulation models corresponding to multiple defects, determine the compositional characteristics of the insulation material decomposition products corresponding to each of the multiple defects, including: establishing a chemical reaction model based on the insulation material in the ring main unit; simulating environmental changes in the ring main unit under multiple defects based on the chemical reaction model and the simulation models corresponding to multiple defects, and determining the composition and concentration of the insulation material decomposition products corresponding to each of the multiple defects at multiple times; establishing a time series of concentration changes corresponding to each of the multiple defects based on the composition and concentration of the insulation material decomposition products corresponding to each of the multiple defects at multiple times; and determining the compositional characteristics of the insulation material decomposition products corresponding to each of the multiple defects based on the time series of concentration changes corresponding to each of the multiple defects.

[0101] Optionally, the processor may also execute program code for the following steps: training a preset enhanced neural network model using the compositional features of the insulation material decomposition products corresponding to multiple defects to obtain a target recognition model, including: determining a training set based on the compositional features of the insulation material decomposition products corresponding to multiple defects, wherein the training set includes the compositional features of the sample insulation material decomposition products and the actual defect types corresponding to the sample insulation material decomposition products; randomly initializing the classifier weights in the preset enhanced neural network model; inputting the training set into the preset enhanced neural network model to obtain the predicted defect types corresponding to the sample insulation material decomposition products; and adjusting the classifier weights of the preset enhanced neural network model based on the predicted defect types and the actual defect types corresponding to the sample insulation material decomposition products to obtain the target recognition model.

[0102] Optionally, the processor may also execute program code for the following steps: inputting the target component and its corresponding concentration into the target identification model to determine the target defect corresponding to the target substation, including: determining multiple combinations of insulation material decomposition products and their corresponding actual defect types based on the insulation material decomposition products corresponding to each defect; inputting the compositional features of the multiple combinations of insulation material decomposition products into the target identification model to obtain the predicted defect types corresponding to each of the multiple combinations of insulation material decomposition products; selecting the combination of insulation material decomposition products whose predicted defect type matches the actual defect type as the target sample combination; and inputting the component and its corresponding concentration that match the insulation material decomposition product type in the target sample combination into the target identification model to determine the target defect corresponding to the target substation.

[0103] Optionally, the processor may also execute program code for the following steps: determining the target defect degree corresponding to the target defect based on the concentration of the target component; determining the target maintenance method based on a preset correspondence, according to the target defect and the target defect degree, wherein the preset correspondence represents the correspondence between the multiple defect degrees corresponding to multiple defects and the maintenance methods.

[0104] This invention provides a method for detecting defects in substation rooms based on the decomposition products of insulating materials. The method establishes simulation models for multiple defects in the ring main unit within the substation room, including partial discharge defects in metal components, surface discharge defects in insulators, air gap discharge defects in bushings, and overheating defects in insulating materials. Based on these simulation models, the compositional characteristics of the decomposition products of the insulating materials corresponding to each defect are determined. A pre-defined enhanced neural network model is trained using these compositional characteristics to obtain a target recognition model. The target components and their corresponding concentrations in the target substation room are detected. These components and their corresponding concentrations are then input into the target recognition model to determine the target defects in the target substation room. This method achieves the goal of identifying defects based on the decomposition products of insulating materials, thereby improving the accuracy of defect type identification. It also solves the problem that current partial discharge detection methods based on electromagnetic and acoustic principles suffer from low accuracy due to complex electromagnetic or acoustic source interference paths in the field.

[0105] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0106] Embodiments of the present invention also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the substation room defect detection method based on the decomposition products of insulating materials provided in the above embodiments.

[0107] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0108] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: establishing simulation models corresponding to multiple defects of the ring main unit in the substation, wherein the multiple defects include partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials; based on the simulation models corresponding to the multiple defects, determining the compositional characteristics of the decomposition products of the insulating materials corresponding to the multiple defects; training a preset enhanced neural network model using the compositional characteristics of the decomposition products of the insulating materials corresponding to the multiple defects to obtain a target recognition model; detecting the target components and their corresponding concentrations in the target substation; inputting the target components and their corresponding concentrations into the target recognition model to determine the target defects corresponding to the target substation.

[0109] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: establishing simulation models corresponding to multiple defects of the ring main unit in the substation, including: establishing an initial simulation model corresponding to the ring main unit; determining the defect location and nature of each of the multiple defects; setting model parameters corresponding to each of the multiple defects based on the defect location and nature of each of the multiple defects; adjusting the initial simulation model based on the model parameters corresponding to each of the multiple defects, and determining the simulation model corresponding to each of the multiple defects.

[0110] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining the compositional characteristics of the insulation material decomposition products corresponding to each of the multiple defects based on simulation models corresponding to each of the multiple defects, including: establishing a chemical reaction model based on the insulation material in the ring main unit; simulating environmental changes in the ring main unit under multiple defects based on the chemical reaction model and the simulation models corresponding to each of the multiple defects, and determining the composition and concentration of the insulation material decomposition products corresponding to each of the multiple defects at multiple times; establishing a time series of concentration changes corresponding to each of the multiple defects based on the composition and concentration of the insulation material decomposition products corresponding to each of the multiple defects at multiple times; and determining the compositional characteristics of the insulation material decomposition products corresponding to each of the multiple defects based on the time series of concentration changes corresponding to each of the multiple defects.

[0111] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: training a preset enhanced neural network model using the compositional features of the insulation material decomposition products corresponding to multiple defects to obtain a target recognition model, including: determining a training set based on the compositional features of the insulation material decomposition products corresponding to multiple defects, wherein the training set includes the compositional features of the sample insulation material decomposition products and the actual defect types corresponding to the sample insulation material decomposition products; randomly initializing the classifier weights in the preset enhanced neural network model; inputting the training set into the preset enhanced neural network model to obtain the predicted defect types corresponding to the sample insulation material decomposition products; and adjusting the classifier weights of the preset enhanced neural network model based on the predicted defect types and the actual defect types corresponding to the sample insulation material decomposition products to obtain the target recognition model.

[0112] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: inputting the target component and its corresponding concentration into the target identification model to determine the target defect corresponding to the target substation, including: determining multiple combinations of sample insulation material decomposition products and their corresponding actual defect types based on the insulation material decomposition products corresponding to each of the multiple defects; inputting the compositional features corresponding to the multiple combinations of sample insulation material decomposition products into the target identification model sequentially to obtain the predicted defect types corresponding to each of the multiple combinations of sample insulation material decomposition products; selecting the sample combinations of insulation material decomposition products whose predicted defect types match the actual defect types as target sample combinations; and inputting the target component and its corresponding concentration that match the insulation material decomposition product types in the target sample combinations into the target identification model to determine the target defect corresponding to the target substation.

[0113] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: determining the target defect degree corresponding to the target defect based on the concentration corresponding to the target component; determining the target operation and maintenance method based on a preset correspondence relationship, according to the target defect and the target defect degree, wherein the preset correspondence relationship characterizes the correspondence between the multiple defect degrees corresponding to multiple defects and the operation and maintenance methods.

[0114] Embodiments of the present invention also provide a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can: establish simulation models corresponding to multiple defects of the ring main unit in the substation room, wherein the multiple defects include partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials; based on the simulation models corresponding to the multiple defects, determine the compositional characteristics of the decomposition products of the insulating materials corresponding to the multiple defects; train a preset enhanced neural network model using the compositional characteristics of the decomposition products of the insulating materials corresponding to the multiple defects to obtain a target recognition model; detect the target components and their corresponding concentrations in the target substation room; input the target components and their corresponding concentrations into the target recognition model to determine the target defects corresponding to the target substation room.

[0115] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0116] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0117] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0118] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0119] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0120] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0121] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting defects in substation rooms based on the decomposition products of insulating materials, characterized in that, include: Establish simulation models for various defects of the ring main unit in the substation room, including partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials. Based on the simulation models corresponding to each of the multiple defects, the compositional characteristics of the decomposition products of the insulating material corresponding to each of the multiple defects are determined. The target recognition model is obtained by training a preset enhanced neural network model using the compositional features of the decomposition products of the insulating material corresponding to each of the multiple defects. Detect the target component and the corresponding concentration of the target component in the target substation room; The target component and its corresponding concentration are input into the target identification model to determine the target defect corresponding to the target substation room.

2. The method according to claim 1, characterized in that, The simulation models established for the various defects of the ring main unit in the substation include: Establish the initial simulation model corresponding to the ring main unit; Determine the location and nature of each of the multiple defects; Based on the defect location and properties of each of the multiple defects, the model parameters corresponding to each of the multiple defects are set; Based on the model parameters corresponding to each of the multiple defects, the initial simulation model is adjusted to determine the simulation model corresponding to each of the multiple defects.

3. The method according to claim 1, characterized in that, The step of determining the compositional characteristics of the insulation material decomposition products corresponding to each of the multiple defects based on the simulation models corresponding to each of the multiple defects includes: A chemical reaction model is established based on the insulation material in the ring main unit. Based on the chemical reaction model and the simulation models corresponding to the various defects, the environmental changes in the ring main unit under the various defects are simulated to determine the composition and concentration of the decomposition products of the insulation material at various times corresponding to the various defects. Based on the composition and concentration of the decomposition products of the insulating material corresponding to each of the multiple defects at multiple times, a time series of concentration changes corresponding to each of the multiple defects is established. Based on the time series of concentration changes corresponding to each of the multiple defects, the compositional characteristics of the decomposition products of the insulating material corresponding to each of the multiple defects are determined.

4. The method according to claim 1, characterized in that, The step of training a preset enhanced neural network model using the compositional features of the insulation material decomposition products corresponding to each of the multiple defects to obtain a target recognition model includes: Based on the compositional characteristics of the decomposition products of the insulation materials corresponding to each of the multiple defects, a training set is determined, wherein the training set includes the compositional characteristics of the decomposition products of the sample insulation materials and the actual defect types corresponding to the decomposition products of the sample insulation materials. Randomly initialize the classifier weights in the preset enhanced neural network model; The training set is input into the preset enhanced neural network model to obtain the predicted defect type corresponding to the decomposition products of the sample insulation material; Based on the predicted defect type corresponding to the decomposition products of the sample insulation material and the actual defect type corresponding to the decomposition products of the sample insulation material, the classifier weights of the preset enhanced neural network model are adjusted to obtain the target recognition model.

5. The method according to claim 1, characterized in that, The step of inputting the target component and its corresponding concentration into the target identification model to determine the target defect corresponding to the target substation includes: Based on the insulation material decomposition products corresponding to each of the multiple defects, determine the combination of multiple sample insulation material decomposition products and the actual defect type corresponding to each of the multiple sample insulation material combination. The compositional features corresponding to the combinations of decomposition products of the multiple sample insulation materials are sequentially input into the target recognition model to obtain the predicted defect type corresponding to each combination of decomposition products of the multiple sample insulation materials. The sample insulation material decomposition product combination that matches the predicted defect type with the actual defect type among the multiple sample insulation material decomposition product combinations is selected as the target sample combination; The target components and their corresponding concentrations that match the type of insulation material decomposition products in the target sample combination are selected and input into the target identification model to determine the target defect corresponding to the target substation room.

6. The method according to any one of claims 1 to 5, characterized in that, Also includes: Based on the concentration of the target component, the degree of the target defect corresponding to the target defect is determined; Based on a preset correspondence, a target operation and maintenance method is determined according to the target defect and the degree of the target defect, wherein the preset correspondence represents the correspondence between the degree of each of the multiple defects and the operation and maintenance method.

7. A defect detection device for substation rooms based on the decomposition products of insulating materials, characterized in that, include: A module is established to create simulation models for various defects of the ring main unit in the substation room. These defects include partial discharge defects of metal components, surface discharge defects of insulators, air gap discharge defects of bushings, and overheating defects of insulating materials. The determination module is used to determine the compositional characteristics of the decomposition products of the insulating material corresponding to each of the multiple defects based on the simulation models corresponding to each of the multiple defects. The training module is used to train a preset enhanced neural network model using the compositional features of the decomposition products of the insulating material corresponding to each of the multiple defects, so as to obtain a target recognition model. The detection module is used to detect the target components and their corresponding concentrations in the target substation room. The identification module is used to input the target component and the concentration corresponding to the target component into the target identification model to determine the target defect corresponding to the target substation room.

8. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the non-volatile storage medium to perform the substation room defect detection method based on the decomposition products of insulating materials as described in any one of claims 1 to 6.

9. A computer device, characterized in that, include: Memory and processor The memory stores computer programs; The processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, the processor performs the substation room defect detection method based on the decomposition products of insulating materials as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for detecting defects in a substation based on the decomposition products of insulating materials as described in any one of claims 1 to 6.

Images